Molluscan Studies Journal of Molluscan Studies (2017) 83: doi: /mollus/eyw042 Advance Access publication date: 4 December 2016

Transcription

1 Journal of Molluscan Studies Journal of Molluscan Studies (207) 83: doi:0.093/mollus/eyw042 Advance Access publication date: 4 December 206 The Malacological Society of London Life cycles and adult sizes of five co-occurring species of Arion slugs John M.C. Hutchinson,2,3, Heike Reise and Grita Skujienė 4 Senckenberg Museum of Natural History Görlitz, Am Museum, Görlitz, Germany; 2 School of Biological Sciences, University of Bristol, 24 Tyndall Avenue, Bristol BS8 TH, UK; 3 Center for Adaptive Behavior and Cognition, Max Planck Institute for Human Development, Lentzeallee 94, 495 Berlin, Germany; and 4 Department of Zoology, Vilnius University, JGMC, Saulėtekio al. 7, Vilnius LT-0222, Lithuania Correspondence: J.M.C. Hutchinson; majmch@googl .com/john.hutchinson@senckenberg.de (Received 2 May 206; editorial decision 8 October 206) ABSTRACT Five species of Arion slugs were collected repeatedly at a woodland site in southern England and all individuals weighed. Selected samples of these were dissected so as to weigh components of the reproductive tract. The relative weights of the gonad, spermoviduct and albumen gland provided the basis to categorize individuals into adult, subadult or immature classes, or as juvenile if the sum of these weights was below a threshold. This procedure was validated by raising A. subfuscus in captivity and killing at a range of known ages before and after egg laying. In the other species, organ weights from individuals observed to have laid eggs or mated also helped to calibrate the divisions. Such data from two species demonstrated that, following the production of an egg clutch, the albumen gland took days gradually to regrow. There was little evidence of much variation in life cycle from year to year and the broad patterns, although not precise timings, agreed with studies elsewhere. No species produced more than one generation per year and in all there was a season (brief in A. subfuscus) when adults were absent. The life cycles were predominantly annual, although in some species a minority of individuals might take 8 months to mature. The time of year at which individuals matured into adults varied between species: A. intermedius in August and September, A. distinctus mostly in December and January, A. circumscriptus mostly January to April, A. subfuscus April to early October and A. rufus July to September. The largest two species thus dominated in summer, but at other times the species overlapped considerably in size. In four species, individuals maturing later in the season did so at a smaller size; the possible exception was A. intermedius, in which maturation was highly synchronized. The coefficients of variation in adult size were compared against a collection of such data from other terrestrial molluscs. The smallest species, A. intermedius, had disproportionately large hatchlings. INTRODUCTION Arion is a genus of slugs native to Europe, attaining its highest diversity in the Iberian Peninsula (Quinteiro et al., 2005). Some species have colonized other continents (e.g. Chichester & Getz, 969) and the genus includes serious agricultural and horticultural pests (Barker, 2002). Our study is a comparison of the life cycles of five Arion species co-occurring at one woodland site. In similar habitats elsewhere in southern England further congeners such as A. hortensis and A. flagellus may also occur, so obvious questions are how these congeners can coexist and whether their niches might be differentiated to reduce interspecific competition. Faecal analysis of the five Arion species that we studied (or of close relatives) has indicated considerable similarities in their diets, predominantly dead higher plant material and underground plant parts (Jennings & Barkham, 975a). Evidence from the wild of intraspecific and interspecific competition (or interference) between terrestrial molluscs is rather sporadic (e.g. Bloch & Willig, 2009; Kimura & Chiba 200), but these processes seem likely explanations for the rapid and widespread replacement in central Europe of A. rufus by the invasive A. lusitanicus auct. non Mabille (e.g. Kappes & Kobialka, 2009). If competition is important, one way that similar species might divide up niche space is by diversifying the timing of their annual cycles. But the effectiveness of such a division is complicated by the wide range of adult sizes within Arion andanevenwider ontogenetic variation in size over the lifetime of each individual (Abeloos, 944). Thus, two species might mature at quite different times of year, yet at some times of year the juvenile slugs of the larger species may match in size the adults of the smaller species. It is unclear how the species could best diversify their life cycles to avoid competition. On the other hand, seasonal stressors such as frosts in winter and drought in summer may exert similar selection pressures on all species, which could lead to synchronous rather than diversified life cycles. Or, the smaller species may be more, or less, sensitive to a seasonal stressor than larger species, leading to some regular relationship between size and the timing of the life cycle. The Author 206. Published by Oxford University Press on behalf of The Malacological Society of London, all rights reserved. For Permissions, please journals.permissions@oup.com

2 LIFE CYCLES OF FIVE ARION SLUGS Life cycles of some Arion species have already been investigated in considerable depth, through regular field sampling and dissection of the slugs found (e.g. Bett, 960; Smith, 966; Hunter, 968; South, 989; Barker, 99). Among these studies, most have concerned agricultural or otherwise non-natural environments, but there are a number of quantitative studies in woodland (e.g. Jennings & Barkham, 975b; Bless, 977; Phillipson, 983; Corsmann, 990). However, few studies have examined more than one, or rarely two, Arion species at the same site. The prominent exceptions are Barnes & Weil s (944) study of four species in suburban gardens and Jennings & Barkham s (975b) study of five species in a deciduous woodland; unfortunately, their reports of the life cycles are just brief verbal descriptions. In contrast, we present our data graphically so that the temporal changes in size and maturity can readily be compared quantitatively across all five species. We are concerned also to compare our results with these earlier studies to examine the consistency of the life cycles across habitats and locations. Complementary information is available from studies of the development and reproduction of several Arion species in captivity (e.g. Künkel, 908, 96; Abeloos, 944; Lūsis, 96; Parivar, 978; South, 982). We use this approach both to calibrate our measures of maturity and to assess the size and growth rate of hatchlings. However, we will uncover several instances in which patterns of growth and maturation of slugs provided with comfortable conditions in captivity diverge from those in the wild; it is not a substitute for studying wild populations. A problem with observing the life cycle of slugs is that adult individuals are usually externally indistinguishable from immature ones. Various studies have used thin sections of the gonad to establish its state of cytological development and in particular whether it was producing spermatozoa and oocytes (e.g. Perrot, 939; Lūsis, 96; Smith, 966; Parivar, 978; South, 989; Barker, 99). This is laborious, so others have taken the shortcut of checking for sperm in the ductus hermaphroditicus as an indication of maturity; sometimes they have also inspected the bursa copulatrix for spermatophores or sperm as evidence of mating (e.g. Bett, 960; Hunter, 968; South, 989). The latter is not appropriate for the two of our species that predominantly selffertilize (Reise et al., 200; Geenen et al., 2006); another limitation is that Arion spermatophores are completely digested within a couple of days of mating (Sionek & Kozłowski, 200; in other pulmonates sperm also disappear from the bursa copulatrix in this timeframe: Dillen, Jordaens & Backeljau, 2009). How closely the filling of the ductus hermaphroditicus matches the transition to being capable of reproduction seems to have been investigated only indirectly (Smith, 966) and one worries that it might look empty again if a recent mating has depleted the sperm. More critically, the presence of sperm in the ductus hermaphroditicus has been used as an all-or-nothing score of maturity, whereas we wanted a more continuous index that could distinguish which animals were in the process of maturing, as well as those that were immature or fully adult. Thus we developed an alternative approach of weighing components of the genitalia so as to compare their proportions quantitatively. This enabled us to uncover some seasonal patterns in adult size overlooked in earlier studies. MATERIAL AND METHODS Sampling The field site was a strip of secondary deciduous woodland, mostly hawthorn, ash and sycamore (Crataegus monogyna, Fraxinus excelsior and Acer pseudoplatanus), with some oak, field maple and hazel (Quercus sp., Acer campestre and Corylus avellana). It lies beside the eastbound motorway service station at Leigh Delamere in Wiltshire, southern England (5.52 N, W). The climate is mild humid temperate, with frequent frosts in winter (December February) and usually periods of one or two weeks in summer (June August) when the litter layer and soil surface largely dry out. Climate statistics from a nearby weather station are provided in the Supplementary Material. For almost 2 years (late 998 to September 2000), we sampled Arion slugs roughly every 2 weeks. Samples from these 56 visits were supplemented by those from 23 sporadic visits in later years (until February 206). The five species collected were, in increasing order of size, A. intermedius Normand, 852, A. distinctus Mabille, 868, A. circumscriptus Johnston, 828, A. subfuscus (Draparnaud, 805) and A. rufus (Linnaeus, 758). Total catch per species ranged from 77 to 963 individuals. Sampling relied on hand searching, always by J.M.C.H., and followed a regular circuit. This included a shallow drainage ditch and various discarded pieces of board and carpet, the undersides of which were checked consistently, but we made a point of also searching in leaf litter. The circuit was c. 00 m long and typically took min. Effort was only roughly similar from visit to visit. Although the full circuit was always covered, searching continued longer and with more stops when slugs were harder to find: the aim was to obtain a sufficient sample to reveal the size distribution, not to estimate abundance. We expect a bias against finding small slugs, but presumably this is a smooth monotonic relationship, so that peaks and gaps in the size distribution should be preserved. Dry weather and low light levels were liable not only to reduce the catch, but also to increase this bias against smaller size classes, which we could aspire only to minimize by looking thoroughly in each spot that was searched at all. Mostly every Arion seen was collected, but not eggs. One exception was that the study started on 5 August 998 with only A. intermedius and the other species were progressively added until 5 December 998, while A. rufus was included only after 3 March To compensate, a few of the collections in later years targeted A. rufus exclusively. Another exception was that occasionally when one species was particularly common or appeared concentrated under few hiding places, we started collecting strictly only every second or every third specimen seen. One potential concern is that our fortnightly collections might have disrupted the age structure of the remaining local population and thus of subsequent samples. However, a 2-week interval provided a reasonable opportunity for slugs in the adjacent woodland to wander onto the narrow sampling circuit. Furthermore, size distributions and densities were similar in later years of the study when sampling was infrequent. Hand searching rarely finds the smallest individuals. To address this weakness, late in the study we collected eggs from the study site, killed some of the hatchlings and reared others for a few weeks in the laboratory until the species identity was clear (A. intermedius, A. subfuscus and A. rufus). For four species (all except A. rufus), we also obtained eggs from specimens collected from the study site when of adult size and kept alive in the laboratory. (These few individuals were weighed alive shortly after collection so as to estimate their preserved mass had they been killed at that point.) We monitored how long the eggs took to hatch, killed some of the hatchlings and reared others for a few weeks, killing a sample occasionally. Conditions in the laboratory did not match those in the wild (see below) but nevertheless the results provide information on hatchling size, a rough estimate of time till hatching (known to be temperature dependent: Hunter, 968) and a probable upper limit for growth rate. Sample processing and the assessment of size and maturity Wild-collected slugs were normally killed within 3 h of collection. Killing of slugs, both wild collected and captive, followed the same method, intended to leave them in an extended state suitable for dissection and without their genitalia protruded. Slugs were placed 89

3 J. M. C. HUTCHINSON ET AL. in warm water (c. 45 C) for a few minutes, then 70% ethanol was added progressively over the course of about h. The ethanol was replaced at least twice over the subsequent month. After some months or years preserved in 70% ethanol (saturated with borax), slugs were weighed to provide our measure of size. Dry weights would not have allowed subsequent dissection, so instead we applied a protocol to standardize how much liquid was removed from the body surface prior to weighing. Each slug was gently rolled using forceps across a consistent type of paper, avoiding squeezing with downward pressure. This continued until almost no wet marks appeared on the paper as the slug was rolled (marks would still appear if it lay still on the paper); this typically took about min. The body was then transferred to a balance (Ohaus AS20; repeatability SD = 0. mg) and the weight recorded after a standard interval (counting to five). Body masses using this protocol averaged 0.7 of the live weight (SE = a factor of 0.08, N = 53) and 5.6 times the weight after drying to a constant weight in an oven at 50 C (SE = 0.6, N = 23). These wet weights of preserved slugs correlate more closely with dry weights than do live weights. For each time of year, we dissected several individuals of different sizes so as to establish at what sizes the state of maturity switched. We dissected relatively few small individuals, since below some size all would be immature, and disproportionately many adult individuals, as part of other research on genital allometry. The entire genital tract was dissected out and weighed using the same procedure as for whole slugs. It was then cut into four parts: the distal genitalia (i.e. nearest the genital pore) and three parts making up the proximal genitalia (i.e. the spermoviduct, the albumen gland, and the gonad together with the ductus hermaphroditicus). These parts were weighed separately, if not too small for our balance. Species identities These dissections also allowed confirmation of the identifications. None of the dissected A. rufus were A. ater, A. lusitanicus auct. non Mabille or A. flagellus. A small minority of the slugs that we have included in A. circumscriptus were A. silvaticus according to the pigmentation of the epiphallus, but these species are not morphologically distinguishable except by colour and may best be considered conspecific (Jordaens et al., 2002; Geenen, Jordaens & Backeljau, 2006; Rowson et al. 204). For those dissected A. distinctus having a tripartite oviduct, we examined the epiphallus structure (Davies, 977). All 79 were confirmed to be A. distinctus except for one A. hortensis collected on a very early visit when the area sampled was more extensive than subsequently. The gonadal characters distinguishing A. subfuscus from A. fuscus (Pinceel et al., 2004) were published after many of the slugs had been dissected, but all adults dissected subsequently fitted A. subfuscus. Furthermore, the 6S rrna gene sequences of six individuals (selected to be representative of a range of seasons and including one of the captive-reared siblings on which Fig. is based) were identical and matched the S haplotype of A. subfuscus (Pinceel et al., 2004; Genbankaccession KX3472). Laboratory culture Preliminary analysis suggested that the proportions of the components of the proximal genitalia could be used to assess maturity (cf. Sokolove & McCrone, 978; Barker, 99; Lush, 2007). To confirm this, we reared three successive clutches from the same mother (an A. subfuscus from the usual field site), killing samples periodically until after some had laid eggs. These slugs were dissected using the same method as for the wild-collected slugs with the additional step that, after weighing the gonad and ductus hermaphroditicus together, we weighed the gonad separately. Also for the other species (except A. rufus) we isolated wildcollected individuals in the laboratory, observed when they laid eggs and killed them a variable number of days afterwards. Arion distinctus was the only one of these four species that mated in the laboratory; some individuals were killed the day after mating (allowing the distal genitalia to retract fully), but before egg laying. A few individuals observed egg laying or mating in the wild also were of use in indicating the anatomy of individuals confirmed to be reproductively mature. Eggs were kept on damp absorbent paper at 2 5 C inthe dark. They were inspected every few days and any hatchlings transferred to a Petri dish containing food. Slugs in captivity were all kept in Petri dishes in a constant temperature chamber (2 5 C) with a circadian light regime. Each Petri dish contained a folded piece of damp absorbent paper, rolled oats, broken cat-food pellet, lettuce, a slice of carrot and a fallen beech leaf. Slugs were transferred to a fresh Petri dish weekly. As slugs grew, we decreased the numbers of individuals in each Petri dish appropriately. Statistical analysis At several points, our analysis examines the relationship between slug size and calendar date or age, or between an index of anatomy and time interval. Because there is no expectation that these relationships are linear, we use Kendall s rank correlation statistic τ, calculated using Wessa (202). RESULTS WITH DISCUSSION FOR EACH SPECIES First, we present results from the Arion subfuscus reared from eggs in the laboratory. Their analysis, complemented by data from A. distinctus, is used to validate and calibrate our method of assessing maturity of wild-collected slugs. Then, for each of the five species of Arion in turn, we present the data from the wild-collected slugs and deduce the life cycle. To avoid repetition and confusion, for each species we follow our conclusions directly with a comparison with the life cycles that others have reported for that species. The final section is a General Discussion, focusing on interspecific comparisons and broader conclusions. Measuring the anatomical transition to adulthood in captive slugs The A. subfuscus reared in captivity from hatching showed a smooth increase in mass with age, with the relative growth rate gradually decreasing throughout the lifetime (Fig. A). There is no clear indication of three distinct phases diminishing in growth rate, as described by Abeloos (944) and South (982) in several Arion species. Little should be made of this disagreement since our data do not cover the first 67 d, the time of the fastest growth phase, and none of the experiments properly controlled for the effect of size on crowding, which is known to affect growth rate (Dan & Bailey, 982; Pearce, 997). More pertinent is that the smooth increase in body weight in our experiment contrasts with the allometric relationships of the component parts of the genitalia. Most parts the distal genitalia, the spermoviduct, the albumen gland and the ductus hermaphroditicus showed a consistent pattern of a sudden increase in size, preceded and followed by much slower increases (Fig. B E). We consider that the step increase corresponds to maturation into an adult. A mixture of animals from both sides of this transition occurred only over the narrow age range of d after hatching; all five individuals killed at 38 d were pretransition and no such animals were observed among the 6 animals killed on or after 77 d. This pattern contrasts with the development of the gonad, which shows a monotonic increase up to the age of about 80 d 90

4 LIFE CYCLES OF FIVE ARION SLUGS A: body B: distal genitalia 0. 0 Preserved mass (mg) C: spermoviduct 0. D: albumen gland E: ductus hermaphroditicus F: gonad Days since hatching Key: juvenile, immature, subadult, adult, undissected, egg layer Figure. Growth of Arion subfuscus in captivity. Offspring from one mother were killed at various ages and components of their genitalia weighed. Symbols indicating scoring of maturity are as derived from Figure 2A. Black circles indicate an individual observed to have laid eggs; some individuals without such a mark may nevertheless have laid eggs (if eggs appeared in a Petri dish containing more than one individual). (Fig. F); older animals all had gonads of about the same absolute size, which is smaller (absolutely) than that of those younger animals that we will designate below as subadult. (This shrinkage of the gonad is known in several Arion species: Künkel, 908; Abeloos, 944; Smith, 966; South, 989; Barker, 99.) Consequently, while immature animals have a gonad larger than spermoviduct and albumen gland combined, in adults these proportions are reversed. For A. subfuscus, based on Figure, we apply an heuristic definition of adults as having spermoviduct and albumen together weighing 0.7 or more of the total mass of the proximal genitalia. By the term adult we imply only that they have fully mature anatomy but, reassuringly, all 2 captive animals observed to have laid eggs met this criterion. If this ratio is less than 0.3, we consider the slug to be immature. Intermediates are termed subadult. (Photographs of immature, subadult and adult genitalia are presented in the Supplementary Material.) These subadults have distal genitalia as large as some adults (Fig. B) so, although they apparently do not lay eggs, they might be capable of exchanging sperm. For our subsequent analysis that issue is not important, but only that subadult is an intermediate maturity class between immature and adult. We also specify a fourth category, juvenile animals whose genitalia weigh less than 3 mg; measurement errors in weighing the component parts of these small genitalia mean that the weight ratios are unreliable. 9

5 J. M. C. HUTCHINSON ET AL. The triangular plots in Figure 2 display the proportions of the proximal genitalia. Here we lump the ductus hermaphroditicus with the gonad, because when measuring the wild slugs we did not separate them. This was a mistake since the two organs show different timings of growth (Fig. E, F). Their combined weight is predominantly determined by the gonad (except in juvenile slugs, gonad weight is at least four times that of the ductus hermaphroditicus, median = 7 times). Position in the state space of Figure 2 shows appreciable variation within the adult class. This can largely be explained by a relationship with time since the last large egg clutch was laid. The albumen glands grew over four-fold in the 0 d or so following the laying of a clutch (Kendall s τ 3 = 0.75, P = , Fig. 3A), whereas the increase in the spermoviduct was not statistically significant and the gonad slightly decreased (τ 3 = 0.55, P = 0.02; the latter might rather be an effect of age, which inevitably A: captive A. subfuscus A B: wild A. subfuscus G Key: juvenile immature subadult adult egg layer mated S C: A. circumscriptus D: A. distinctus E: A. intermedius F: A. rufus Figure 2. The proportional masses of components of proximal genitalia of Arion species. Slugs in which gonad (with ductus hermaphroditicus) predominates appear near corner G; in those near corner A albumen gland predominates; in those near corner S spermoviduct predominates. Centre of triangle corresponds to an equal mass of all three components. Symbols indicating maturity are defined on basis of position in these diagrams (dashed lines indicate boundaries), except that if all components together do not exceed a critical value the slug is defined as juvenile. A black circle indicates a slug that has produced eggs, a black cross (in D only) one observed to mate. These latter two symbols apply mostly to slugs in captivity; otherwise coloured symbols in B F show only wild-collected slugs. A shows captive-reared offspring of one mother (as in Fig. ). 92

6 LIFE CYCLES OF FIVE ARION SLUGS Albumen gland/(albumen gland + spermoviduct) A: A. subfuscus B: A. distinctus Days since last big clutch Figure 3. The albumen gland regrows following laying of a large clutch. Mass of albumen gland is shown as a fraction of combined masses of albumen gland and spermoviduct. For Arion distinctus (B), the experiment was repeated with three groups, distinguished with different symbols; with the largest group the increase is highly significant, one repetition shows same trend (but with only four individuals), but there is no effect in third group; overall the effect remains significant (τ 25 = 0.42, P = 0.004), as it is with A. subfuscus (τ 3 = 0.68, P = 0.002). correlates with time since the last clutch). We conclude that much of the mass of the albumen gland is transferred into eggs as they are laid, following which the albumen gland gradually regrows in preparation for the next clutch. Others have previously drawn a similar conclusion simply from the variability in the size of albumen glands among adults (Tompa, 984; Tomiyama, 993), but ours is a more direct demonstration. Slugs killed after egg laying appear displaced slightly lower in Figure 2A than those growing their albumen gland for the first time. Adult slugs continued to grow after they matured (Fig. A: τ 8 = 0.46, P = 0.0); the reason is not just the periodic swelling of the albumen gland, because the increase persists if we subtract albumen-gland mass from body mass (τ 8 = 0.40, P = 0.02). In A. distinctus, again the albumen gland increased with time since the last large clutch (τ 25 = 0.5, P = ; Figs 2D, 3B), but gonad size shows no relationship. In this species we have data also on animals known to have mated recently; five animals collected from the study site were killed after mating in captivity, four animals killed shortly after collection were noted as having remains of spermatophores in their bursa copulatrix, and two pairs of animals from other sites in England were killed after having been found mating in the wild. These mating-capable individuals appear within the cluster of animals observed to have laid eggs (Fig. 2D). In both A. subfuscus and A. distinctus, the triangular plot for wild slugs has the same features as for the laboratory-reared population (e.g. Fig. 2A, B), even though they never grow as large in the wild. We therefore apply the same criteria for defining the maturity classes. In the other species, we have less data to indicate the region of the triangular plot corresponding to egg-producing individuals. Nevertheless, the five species exhibit similar distributions on these plots, so we constructed similar limits between maturity classes by eye largely on the basis of the interspecific analogy. We cannot be sure how precisely our divisions between subadult and adult classes correspond with the onset of mating or egg laying. However, most of our conclusions follow only from the reasonable assumption that those classified as adults are more mature than those classified as subadult, and similarly with the ranking of immature and juvenile categories. Arion subfuscus in the wild Variation in the timing of the life cycle between years would be unsurprising, but our sampling was not appropriate to investigate this quantitatively. We applied the working assumption that this variation was minor and thus for each species have pooled all years together when plotting how size distribution changes with time of year (Fig. 4). The patterns in these plots that we discuss are evidently consistent enough not to have been obscured by interyear variation. If there were anomalies, we did investigate interyear variation as a possible cause. Besides these scattergrams, a series of histograms of the size distribution each month is presented in the Supplementary Material; some patterns stand out more in one style of presentation, some in the other. Table provides a summary of our conclusions about each species. In A. subfuscus, the distribution of size through the year is based on 474 individuals collected, of which 3 were dissected (Fig. 4A). Adults were commonest from June to October but some were found at almost all times of year. The exception is that the preceding year s generation appears to have died off just before the next generation of subadults and adults appeared in March and April respectively. Subadults occurred only from March to October. Over this period, this transitional stage of development occurred in progressively smaller slugs (τ 23 = 0.67, P < 0.000) and the adult slugs also tended to be smaller later in the season (τ 30 = 0.49, P = ). Subadults in March are on average 3.9 times as large as those collected 6 months later. The reduction in adult size over the same period is less steep, a factor of 2.8, which is to be expected since the larger earlier-maturing adults persist in the population. This reduction in adult size is compatible also with the possibility that adults shrink as they age and egg laying will also reduce body mass at least temporarily. However, there is no evidence of a change in size among adults after the last subadult individual was recorded (November to March: τ 5 = 0., P = 0.6). In our captive population, adults continued to grow (Fig. A). However, from April to July adults in the wild are little bigger than the coexisting subadults. After July the adults are larger than the subadults present at that time, but this could be because these adults have maintained their size after maturing from the large subadults present earlier in the season. Our conclusion is that individuals in the wild may grow little or not at all once they become adult. We found an individual laying eggs on 30 August. Also we collected eggs on 30 December, which hatched over 8 29 January. Judging from the rate of growth of the captive hatchlings, such eggs could have generated the distinct cohort of small individuals first found in March and gone on to yield adults that summer. However, as it approached adult size, this cohort merged with a cohort of larger but also immature slugs, which was clearly already existent in March and probably since the previous autumn. One possibility is that these were hatchlings from January of the previous year that did not grow fast enough to become adult in the previous summer and so delayed growth and maturity to the next spring (i.e. a life cycle exceeding year). Another nonexclusive possibility is that adults laid eggs through much of the year, so that there were always a few small individuals developing over autumn and winter. If they postponed further growth and maturation once attaining near-adult size, a cohort of large immature individuals would accumulate over the winter. 93

7 J. M. C. HUTCHINSON ET AL A: A. subfuscus Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Preserved mass (g) B: A. circumscriptus Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan C: A. distinctus Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Collection date Figure 4. Continued on next page 94

8 LIFE CYCLES OF FIVE ARION SLUGS Preserved mass (g) D: A. intermedius Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Collection date Key: juvenile, immature, subadult, adult, undissected, hatchling. E: A. rufus Figure 4. The preserved weights of five species of Arion plotted against day of collection, for all years combined. Undissected slugs appear as small black crosses, other symbols signify maturity as assessed by proximal genitalia (see Fig. 2). Horizontal bars indicate weights of hatchlings (from captive-laid eggs in B E and/or wild-collected eggs in A and D; mean value per clutch in D, otherwise of single individuals) and solid grey lines indicate their growth in captivity. In C, dashed grey line connects median weight of hatchlings to median mass of other slugs reared in crowded conditions, with horizontal displacement corresponding to their age when killed. See also Supplementary Material for alternative presentation in which histograms show size distribution each month. For one particular clutch some eggs hatched after d (we did not check for eggs or hatchlings every day, hence the 4-d range), whereas others under identical conditions took at least 52 d; a second clutch hatched within this time range. Hatchlings from eggs of the large captive-reared mothers were considerably larger than from those found in the wild (median mass of 5 hatchlings from two clutches laid in laboratory = 6.4 mg, range mg; median mass of 8 hatchlings from eggs collected on 30 December 202 = 2.8 mg, range mg). Comparisons with earlier work: Barnes & Weil (944), in a study of gardens north of London, reported a life cycle matching our observations in most respects: almost all individuals in July and August were large; these gradually died off over autumn and winter, with abundance at a minimum from October to early December; small specimens occurred from September until early summer. However, we still observed many small specimens in June, whereas they did not. They found large numbers of newly hatched specimens in winter, especially December to February; we did not, but such a hatching time would fit with when we found eggs and with the appearance in March of slightly larger individuals. If hatchlings hide underground during the day, we would have missed them, whereas Barnes & Weil (944) could have found them because they collected active slugs at night. Bett (960) worked at the same field site. From May to December she dissected slugs, using the criterion of sperm in the ductus hermaphroditicus to define maturity and sperm in the bursa copulatrix to infer past mating. Mature-unmated slugs occurred from 95

9 J. M. C. HUTCHINSON ET AL. Table. Summary of life cycles of five Arion species at Leigh Delamere. A. subfuscus A. circumscriptus A. distinctus A. intermedius A. rufus Period of life cycle Annual Annual + 8 months Annual (?+ 8 months) Annual Annual (?+ biennial) Subadults present March October November April October January + July September June September March April Subadults and adults April October December April December January + August September July September present March April Adults present April March December August December August August April July September Seasonal decline in size Yes Yes Yes NS Yes at which mature? Adults grow or shrink? Little or not Grow then?shrink Uncertain Some growth No evidence Special features Adults almost always present Bimodal size distribution of juveniles in autumn Summer splits cohort; mature at two seasons Highly synchronized; variable survival to spring Short season when adults present The simultaneous presence of subadults and adult indicates when transition between these states is occurring. June to August and mature-mated slugs from July to November; our technique implied maturation 2 months earlier than this. Bett collected egg-laying slugs in August, with these eggs hatching after 6 weeks in an unheated room. This agreed with her finding many young in October, earlier than Barnes & Weil (944) observed; but Bett (960) noted that the hatchlings grew only slowly until spring, so maybe later hatchings in winter could have added to this cohort without attracting notice. Jennings & Barkham (975b), working in woodland in eastern England, similarly found that slugs reached adult size from May to September. Young hatched from August to March with a peak in October and November. Bless (977), sampling in woodland in western Germany, also observed that his size class of the largest slugs showed a strong peak in abundance during June, but this size class was present throughout the year except for the two coldest months. Beyer & Saari (978), sampling a field in New York State, observed the same June peak, but the adult population declined markedly in August so that large slugs were absent around September and October, roughly as we found. However, in contrast to our results, a pulse of small slugs (<4 mm) appeared already in August and September; these formed a clearly delimited cohort maturing the following summer. They commented that other studies, not all based on quantitative fieldwork, varied considerably in when they concluded the peak of hatching occurred. In the adjacent state of Connecticut, Chichester & Getz (973) observed adults somewhat later in the year (late summer to early autumn), with a peak of egg laying in September and October, and the adults had become rare only by November, although occasional individuals seemed out of synchrony. Arion circumscriptus in the wild The triangular plots of the components of the proximal genitalia suggest a developmental trajectory similar to that for A. subfuscus (Fig. 2). By analogy with that species, and taking account of the position of the four captive specimens that laid eggs, we specified developmental stages as shown in Figure 2C. Size distributions through the year derive from 574 individuals, of which were dissected (Fig. 4B). Subadults occurred only between November and April. Adults appeared a little later (January onwards except for one in December) and some survived until July or exceptionally August, implying that adults can survive at least 3 4 months. Weights of subadults correlate negatively with time since November (τ 5 = 0.64, P = 0.00), implying that individuals matured at a smaller size later in the season. Weights of adults show the same decline, even if we consider only adults found after no subadults remained to mature (τ 33 = 0.34, P = 0.008). One explanation would be that the older larger adults were dying off earlier than the younger smaller ones, shifting the mean mass downwards. Another possible explanation is that individual adults were shrinking as they became old. Since the last two adults collected were smaller than any collected earlier, Figure 4B hints that the latter explanation operates. On average subadults found in November were 2.3 times as heavy as those found 5 months later in April; and adults found in January were 2.7 times as heavy as those found 6 months later in July. We conclude that adults initially continued to grow after maturing, because most are considerably heavier than subadults: the heaviest subadult weighed 0.39 g, which 58% of adults exceeded; the heaviest adult was 0.93 g. This contrasts with the apparent lack of adult growth in A. subfuscus. Potentially, such a difference might be an artefact of defining the subadult class differently in the two species, with the subadult class in A. subfuscus really at a more mature stage than subadults of A. circumscriptus. However, adult A. subfuscus are little larger even than larger individuals of the immature class (Fig. 4A). The size distribution in May indicates a cohort of slugs of about 0. g that could not become adult that season (Fig. 4B). They largely disappeared over high summer (August), presumably burrowing underground to escape the dry weather, but reappeared in autumn, grew and became adult a few months later. The origin of this cohort requires some discussion. The most prominent pulse of young slugs appeared in July and August. The initial rate of growth of this cohort is clear and, although rather slower than for slugs in captivity, it seems sufficient to have taken them to adulthood the following spring (i.e. an annual life cycle). Five small slugs (<0.02 g) found in December appear to be slow developers (or late hatchers) of this cohort because this is too early for them to have been offspring of the next generation of adults. Their size and date of occurrence lines them up well to generate the cohort that survives the following summer half-grown and becomes adult at the turn of the year. In that case they would represent individuals with a life cycle of a year and half. There were also a few small slugs appearing from January to March; these are plausibly offspring of the cooccurring adult generation that had started to reach adulthood in December and they would have reached adulthood the following year (again an annual life cycle). Four of the slugs collected on 22 April 202 laid eggs between 6 May and 3 June. Eggs took a minimum of 30 d to hatch, most taking about 34 d, the longest taking d. Hatchlings had a median mass of 4.6 mg (range mg; clutches varied significantly, F 3,35 = 5, P < 0.00, with medians mg). In the wild, we found no very small juveniles (<0.03 g) in April, May or June. As confirmation of the reality of this absence, once the new cohort of hatchlings did appear in August, there was a conspicuous size gap between them and the cohort that survived the summer half-grown. The interruption to recruitment in spring 96

10 LIFE CYCLES OF FIVE ARION SLUGS is unexpected because the parental generation was present as adults continuously from January to July. Perhaps, while some early-maturing adults did lay eggs in December, most adults delayed laying eggs until the end of their lives in mid summer. Or, eggs continued to be laid but those that could hatch in spring either delayed hatching until July and August or produced hatchlings that hid underground without growing until that time. If some of these eggs or hatchlings continued dormant for longer, that would explain the origin of the small slugs appearing in December. The puzzle would best be addressed by collecting eggs early in the year and keeping them in natural conditions to monitor when they hatch. Comparisons with earlier work: Jennings & Barkham (975b) studied the closely related A. fasciatus in English woodland. Their soilextraction method allowed them to find hatchlings in nearly every month, but with peaks in January and August to September; thus they too noticed less recruitment in late spring and early summer. Despite the colder winters, the life cycle in Moscow gardens showed close similiarities with our population (Burenkov, 977). The first juveniles appeared in July and the adult generation died off in August. The population overwintered at a wide range of sizes and matured to reproduce in June and July (i.e. delayed relative to our population, presumably by the Russian winter). Burenkov (977) further suggested that a minority of animals overwinter a second time, as recently matured adults; his evidence for this seems weak but the phenomenon would match what we suspected. In Michigan grassland, Getz (959) commented on the abundance of adults in spring and summer but, puzzlingly, he observed eggs only from September until the time of hibernation. Arion distinctus in the wild We collected 963 animals (approaching twice as many as of any other species) of which 38 have been dissected (for another dozen the internal organs were too parasitized for weighing of genitalia to be meaningful). Animals classified as adults (Fig. 2D) were present from December to August, although from mid-june to August few individuals of any kind were collected, presumably because of the dry weather (Fig. 4C). Subadults were found only from October to mid-april, so that later in spring and in summer the population consisted of two anatomically distinct cohorts, adults and juveniles, overlapping in size. Very small slugs (<5 mg) were collected at the beginning of March but also in July and August; one individual collected on 29 May laid eggs that day. Our interpretation is that adults lay eggs throughout the winter and spring until they die off in summer. Eggs laid early in this season yield slugs that reach the size of small adults before summer, scarcely grow over the summer and grow further through autumn, but become adult only in December or January. Other individuals hatch later in spring and summer, and remain small over the summer; these grow through the autumn, but a strongly bimodal size distribution persists in December and it is subadults from this second, less numerous, cohort that appear in March and April. Between mid-january and mid-march we identified no subadults, only less mature or fully adult slugs. Thus, summer seems to split the population into two cohorts developing in parallel. The cohorts are not genetically isolated because adults from both coexist after March. We cannot rule out that a minority of the second cohort do not grow enough to become adult in April and instead survive a second summer, yielding a life cycle of.5 years. Subadults collected later in the reproductive season tended to be smaller (τ 22 = 0.4, P = 0.0), implying that maturation occurs at a smaller size. In particular, over the weeks either side of New Year, there appears to be a sharp decline in the size at which subadults become adult; the smallest adult in December weighed 0.4 g, whereas adults as small as 0.5 g occurred at the end of January. However, the influence of size on the decision to mature should not be overemphasized. In contrast to A. subfuscus and A. circumscriptus, some cue triggers slugs of a wide range of sizes to become adult within just a few weeks of each other. Subadults of the second cohort becoming adult in March and April were a similar size to subadults in January. Over the whole reproductive season, adult size declined (τ 70 = 0.9, P = 0.02), echoing the pattern with subadults. But if we consider adult size only from mid-april to August, after the last subadults were recorded, the decline in size was not significant (τ 26 = 0.4, P = 0.35), suggesting that adults do not usually shrink as they become older. However, two adults from April and May that were smaller than any subadults found might have been shrunken senile individuals. Abeloos (944) reported such a decline in size in a captive population. Do adults continue to grow after maturing? Adults tended to be a little larger than subadults (median 0.3 vs 0.23 g, P = 0.03, Mann Whitney test), but this could be because of growth in the subadult phase. The difference is no longer significant once we subtract the weights of the albumen gland from those of the body. The largest adult was little bigger than the largest subadult and, as just noted, adult size does not increase with time even after the last adults have matured. So there is reason to doubt whether adults continue to grow after they mature (except for the periodic accumulation of resources in the albumen gland). This species exhibits a dimorphism in the oviduct, a structure that in one morph strokes over the back of the partner during copulation (Davies, 977). We scored 32 individuals as bipartite morphs and 79 as tripartite. Their relative frequencies showed no obvious relationship with time of year or with size at a particular time of year. Twelve hatchlings from a clutch of a wild-collected slug weighed a median of 2.2 mg (range.0 2.7). That clutch started to hatch after d and the last egg hatched at 28 d. Data from another six clutches laid by captive-reared slugs were all compatible with taking 28 3 d to hatch. With this species we did not follow the growth of young slugs in captivity, but slugs reared for a different experiment provide some information. These slugs were kept rather crowded (0 per Petri dish), but nevertheless 95 d after hatching they had grown to adult size ( g). None of the five dissected at this age had adult anatomy. In this experimental population, the shortest interval from hatching to egg laying was 5 months, but some individuals took more than twice as long, apparently lacking some cue to trigger maturation. Comparisons with earlier work: Only in 977 did Davies distinguish A. distinctus from A. hortensis. Articles reporting the life cycle of A. hortensis predated or overlooked this work, so it is uncertain which species they describe, or whether they dealt with mixed populations. Davies (977) herself had observed that in mixed populations A. hortensis was a few weeks or months in advance of A. distinctus. At two sites she found high proportions of freshly mature A. distinctus in November but considered that mostly the species was not ready to breed until the latter part of the winter or even spring. However, she also stated that A. distinctus matures throughout the year. Barnes & Weil (944) in a study in gardens north of London observed that A. hortensis became rare in July, but that individuals of all sizes occurred throughout the year. This would be a reasonable description of our data; it was only our additional information on maturity that revealed the strong seasonal pattern. Bett (960), working at the same field site, inferred maturity from the presence of sperm in the ductus hermaphroditicus (i.e. mature ) and in the bursa copulatrix (i.e. mated ). Her technique indicated that slugs in July and August were immature, in agreement with our results, but that mature-unmated appeared already in September or October and mature-mated in October. This is in advance of when we judged adults to appear (December), which might result merely from a difference in definitions. She did observe the same seasonal decline in mean size of mature 97